[gmx-users] Slow Heating and Cooling

Martina Bertsch, PhD mbe404 at lulu.it.northwestern.edu
Wed Jan 21 10:24:01 CET 2004


The simulated annealing algorithm implemented in Gromacs enables:

1) heating of structures from ref_t K to infinity, if the zero_temp_time 
< 0 and

2) cooling from ref_t to 0 K, if the zero_temp_time > 0.

Other parameters are:
annealing = yes
tau_t, which if increased, allows for a slower heating.

Also, longer simulations imply slower cooling.

But, how exactly do I program a slow cooling from, say, 600 K to 500 K 
(instead of 0 K)? Which mdp options do I use? If I say:

gen_vel = yes
gen_temp = 600
ref_t = 500

wouldn't temperature just quickly decrease to 500 K and remain at 500 K 
for the rest of the simulation? Or could I force a slow temperature 
decrease by increasing tau_t? What kind of value for tau_t would I need 
so that the cooling from 600-500 K of a system of approx. 3000 atoms 
takes 5 ps?

In the gmx archive, I found a discussion on the slow heating from 0 
to100 K, in which Anton suggested setting init_t value to something 
other than zero, but I did not quite understand what he meant. Would he, 
please, elaborate? Is it at all possible to stop the heating at 100 K, 
or to predict when the 0-to-infinity heating will reach 100 K?

I would appreciate your comments.


More information about the gromacs.org_gmx-users mailing list